Table of Contents

Enum OcrUnderline.Trait

Namespace
LMKit.Extraction.Ocr
Assembly
LM-Kit.NET.dll

What the page around a mark says about it, as the engine read it from the pixels.

[Flags]
public enum OcrUnderline.Trait

Fields

None = 0

Nothing beyond the stroke itself.

TextBelow = 1

Ink sits under the stroke within one text height, over as many columns as the text above it: a row separator or a tightly set next line.

BoxEdge = 2

A vertical ink run meets the stroke at one of its ends: a box corner or a cell wall.

Thick = 4

The stroke is thicker than half the height of the text above it.

Pieces = 8

The stroke was read as a row of short pieces (dashes, dots, the rules of single words) rather than as one run.

NotErased = 16

The stroke was reported but left in the image the recognizer read.

Rule = 32

The stroke was a rule the engine's line finder erased.

LeaderRow = 64

The pieces were erased as a leader row before the underline pass.

GridTwin = 128

Another horizontal stroke shares both ends of this one within a few pixels, stands between 0.8 and 4 text heights away with a row of text between the two, and a vertical crosses it: the two are the row rules of a grid.

Faint = 256

The stroke was read on the grey page under the binarizer's threshold, where the binary page carries nothing of it: a hairline or a row of dots whose anti-aliased grey stays lighter than the threshold. A weaker signal than a binary stroke, reported for a consumer to weigh; never erased.

Share